some questions :) - testing mwes
Posted: Fri Nov 03, 2017 8:41 am
My english is bad (sorry):
If I create user and then I disable it, user can still log in. Same problem is, if I add email to account, user can read all emails. mwe v. 3.0.0.17 debian stretch, maybe bug ?
In future (roadmap) is possible train mail via shell scripts (moving mail between maildirs) then our server run script something like "mwelearn --spam/ham --user=email/global content_of_email" (<< EOF ... etc), this work on same machine, but maybe via API to remote mwes ? Our users use this concept to learn spam. Moving email to spam folder learn mail as spam, moving to inbox from spam learn as ham. (Bayes training)
Quarantine is not bad, but UI support only eng lang., I need gettext, xml ... to translate UI for customers. I can translate tmpl files, but this is bad idea for future upgrades.
TLS missing, I read here something about proxy and problems around this ... ok
Another idea - Problem is, when spammers use second MX server for spamming and this server is in whitelist. Actual license cover one server, any idea how to backup secondary MX ? My idea is switch mwes to backup (secondary) mode (without UI), syncing data from master mwes.
Report spam link in users mail bypassing bayes limit in settings ? Or replace data in bayes DB ? Our actual antispam corpus have 3000/3000 spam/ham and 500 samples ... Is that not a little too little?
I use goole translate sorry Thank you for your response.
If I create user and then I disable it, user can still log in. Same problem is, if I add email to account, user can read all emails. mwe v. 3.0.0.17 debian stretch, maybe bug ?
In future (roadmap) is possible train mail via shell scripts (moving mail between maildirs) then our server run script something like "mwelearn --spam/ham --user=email/global content_of_email" (<< EOF ... etc), this work on same machine, but maybe via API to remote mwes ? Our users use this concept to learn spam. Moving email to spam folder learn mail as spam, moving to inbox from spam learn as ham. (Bayes training)
Quarantine is not bad, but UI support only eng lang., I need gettext, xml ... to translate UI for customers. I can translate tmpl files, but this is bad idea for future upgrades.
TLS missing, I read here something about proxy and problems around this ... ok
Another idea - Problem is, when spammers use second MX server for spamming and this server is in whitelist. Actual license cover one server, any idea how to backup secondary MX ? My idea is switch mwes to backup (secondary) mode (without UI), syncing data from master mwes.
Report spam link in users mail bypassing bayes limit in settings ? Or replace data in bayes DB ? Our actual antispam corpus have 3000/3000 spam/ham and 500 samples ... Is that not a little too little?
I use goole translate sorry Thank you for your response.